Studies of Reactions of Atoms in a Discharge Flow Stirred Reactor Part 3.-The 0 + H, + 0, System B Y IAN M. CAMPBELL," JOHN s. ROGERSON AND BRIAN J. HANDYt School of Chemistry, The University, Leeds LS2 9JT Received 12th April, 1978 By addition of H2 to O(3P) atoms in N2 carrier, a partial conversion of 0 atoms to H atoms was achieved through the reactions O+H2 -+ OH+H (1 1 O+OH +02+H (2) in a discharge-flow stirred reactor at 425 K. Addition of small amounts of CO (c 5 %) generated bluish 0 + CO chemiluminescence, the intensity of which at entry and exit ports was measured to establish 0 atom decay rates. When O2 (< 1 %) was added, H02 radicals were formed in situ by reaction H+02+M+HOz+M (9) H+H02 + 20H (10) O+HO2 + OH+02 (11) H+HO2 +H2+02 (22) H+HO2 H20+0 (23) and the 0 atom decay rate was accelerated due to subsequent reactions and reaction (2).The parallel steps to reaction (10) consume H atoms and reaction (23) produces 0 atoms so decreasing the catalytic rate of removal of 0 atoms. From the variation of the catalytic rate as a function of [H]/[O] ratios in the range 0.16 to 3.46, a value of k9(M = N2) = (1.2k0.2) x 10" dm6 mok2 s-' at 425 K was obtained, together with a placing of the ratio kll/(k10+k22+k23) (i.e. the relative reactivity of H02 with 0 and H atoms) in the range 0.2 to 0.5 at 425 K. The H02 radical has long been recognized as an important species in combustion and explosion phenomena and more recently as a significant component of strato- spheric chemistry. The reactions of H02 with O(3P) and H(2S) atoms play major roles in such situations.For example Donahue et a2.l have pointed out the critical role of reaction of H02 with O(3P) atoms in determining the depletion effects of chlorofluorocarbons upon the stratospheric ozone layer, but these can only be quantified if rate constants are available for all the elementary reactions concerned. we have described the operation of the discharge-flow stirred reactor, when it was used to synthesize HNO and HCO radicals by in situ three body combination reactions (H + NO/CO + M) and thence to measure their relative reactivities with H(2S) and O(3P) atoms. In this study 0, is added to the 0 + H2 + N2 system to synthesize H02 by combination and thence to establish the ratio of rate ?Present address : Department of Physical Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EP.In Parts 1 and 2,29 2672I . M. CAMPBELL, J . S. ROGERSON AND B. J . HANDY 2673 parameters for this subsequent reactions with H and 0 atoms, which has not been measured directly before. As in the previous work 2 * the method is based upon the catalytic consumption of 0 atoms, here induced by the presence of H02 in the reactor. EXPERIMENTAL The general procedures used in this work were similar to those used previously.2* The central feature was a Pyrex sphere of internal volume 0.54 dm3, internally-coated with syrupy phosphoric acid to inhibit wall recombination of atoms. The entry and exit tubings were non-penetrating and in-line : the entry tubing had inset jets J1 and J2, the latter just upstream of the observation point L1, as shown in fig.1 of Part 1.. Between these two jets was a sidearm through which CO and O2 were added as required. At the upstream jet, Jl, N(4S) atoms in the discharged N2 were titrated with NO, so generating known concentra- tions of O(3P) atoms according to the stoichiometric and rapid reaction represented by the equation N(~S)+NO -+ N~+O(~P). The disappearance of visible emissions, associated with N+ N, N+ 0. or O+ NO combina- tions, in the tubing below J1 marked the equivalence point of the titration. The intensities of the bluish O+ CO chemiluminescence were measured at the observation points L1 and L2, at the entry and exit of the sphere respectively, using an RCA 1P28 photomultiplier viewing through an Oriel Optics G-774-3550 coloured glass filter as b e f ~ r e .~ The photomultiplier signals, displayed on a Pye Scalamp galvanometer, were proportional to [O] and had been intercalibrated to provide the oxygen atom decay parameter { - A[O]/([O]At)) as described b e f ~ r e , ~ where A[O] is the difference between the 0 atom concentrations at entry and exit, [O] here is the uniform 0 atom concentration in the sphere and At is the residence time of the gases in the sphere. Calibrated flowrates of H2 (<lo % of total flowrate) were added through the sidearms of the sphere. This and the other gases were purified and delivered as described Total pressures were in the range 0.2 to 0.5 kPa, measured using a silicone oil manometer balanced against running vacuum. Total flowrates were < 100 pmol s-' and the ratio of residence to diffusion time was always >5 to ensure stirred flow operation.2 The sphere was enclosed in an insulated box fitted with a heater and air circulating arrangement and the temperature used, 425 K, was measured using a thermocouple junction in contact with the external wall of the sphere.RESULTS AND DISCUSSION The temperature of 425 K was chosen since this allowed greatest flexibility in generating [H]/[O] ratios in the range 0.1 to 4 without H2 becoming a more than minor component in the stirred flow reactor. By varying the H2 flowrates in the range 0.4 to 7 pmol s-l, the reactions 0+H2 + OH+H (1) OfOH 3 0 2 f H (2) [with reaction (1) rate determining and pseudo-first order] achieved the partial conversion of O(3P) to H(2S) to an extent dependent only on [HJ.The presence of the small amounts of CO induced slight competition for OH radicals through the reaction (reaction numbering scheme continues from parts 1 and 2).** Our recent measure- ment of k2/kL8 = 260120 at 425 K allowed us to calculate values of a parameter CO+OH 3 COz+H (18)2674 O+H,+O, SYSTEM r = k,[O]/(k,[O]+k18[CO]) used in the analysis since [CO]/[O] was known. cycle of reactions studied in Part 2 The I-I+CO+M 3 HCO+M (19) H+HCO -+ H2+C0 (20) O+HCO 3 CO+QH O+HCO -+ C02+H was of little significance under our present conditions with [CO]/[O] < 40 but a small correction was applied using the values of k19, k20/k21 and k21a/k21b measured previ~usly.~ The reaction HCO+O,-+ CO+HO, has a rate constant only - 3 % of kal additional competition for OH radicals induced by reaction was corrected for on occasions of any significance by incorporation of a term kl2[H2] (21a)}(21) (2 1 b) and was, therefore, insignificant.The small OH+H, -+ H,O+H (12) into the denominator of r as b e f ~ r e . ~ The addition of O2 to the O+H2+N2 the decay parameter { - [AO]/([O]At)} as significant reactions in this respect are H+O,+M 3 H+HO2 3 O+HO2 system leads to a linear enhancement of was shown in fig. 3 of Part 1.2 The with OH radicals reacting in (2) subsequently. possible pathways, the others being Reaction (10) is one of three parallel H+HO2 -+ H2+02 H+H02 -+ H 2 0 + 0 . Reaction (22) exerts no influence on the 0 atom decay rate while reaction (23) will decrease it. But reactions (22) and (23) consume H atoms and so will decrease the [H]/[O] ratio compared to that in the system without 0, addition.The ozone- forming reaction will also occur in the system and will be followed by the reaction However, the rate constant k24 is only around 1 x lo8 dm6 mo1-2 s-l [ref. (6)] as opposed to k9 being around 1 x 1O1O dm6 mo1-2 s-I (see later), both for M = N2 at 425 K. Accordingly since [H]/[O] 0.1 in this work, reaction (24) can only make a minor contribution (< 10 %) to catalysed 0 atom decay in comparison with reaction (9). Thus a small correction can be made based on the known value of kz4 without introducing significant uncertainty. Analysis of the above mechanism yields the expression for the decay parameter O+02+M -+ 0 3 + M (24) H+03 4 OH+-,.(25) (ii)1 . M. CAMPBELL, J . S . ROGEKSON AND B . J . HANDY 2675 The subscript c denotes that the decay parameter has been corrected for the small effects of reactions (19), (20), (21) and (24) and G is evidently the gradient of a plot of (- A[O]/[O]At), against [O,]. The [H]/[O] ratio in the reactor is given by after correction for the minor effects of reactions (19)-(21). The [H]/[O] ratio for a particular experimental determination of G was taken as an average calculated by an iterative procedure applied to eqn (iii). Literature values and our own results (see later) suggested kg (M = N,) - 1 x 1O1O dm6 mo1-2 s-I at 425 K. Also existing literature values (see later) suggested 2 kl - klo + k22 +k23. These values were applied first, the resultant [H]/[O] was applied to eqn (ii) to generate an improved set of rate constants for reapplication in eqn (iii).The values of [H]/[O] converged rapidly to a constant value largely because the denominator term in eqn (iii) never exceeded 1.15 under our conditions ; it was particularly close to unity for the most crucial experiments with the lowest [H]/[O] ratios. For the main analysis of results it was useful to define two ratio parameters; RH = (k,o-k,3/2r)/(klo+k2,+k23) and R, = kll/(klo+k22+k23), the latter then expressing the relative reactivity of HO, radicals with 0 and H atoms. Eqn (ii) may then be expressed Under our conditions of [CO]/[O] < 40, the value of r was - 0.9 and thus 1 + Y and 2r are almost equal. The error involved in replacement of both by the average between the two, roughly weighted in one direction or the other depending on the value of [O]/[H), was considerably less than the experimental uncertainty.Thus division of the left hand side of eqn (iv) by this average value (- 1.85) then yields an apparent value of kg denoted by k,(app). From the measurements of G from the apparently linear plots of (- A[O]/[O]At), against [O,], when 0, was < 1 % of the total gases, a set of values of k9(app) was obtained for 0.16 < [H]/[O] < 3.46 and these are plotted in fig. 1. Strong upward curvature is evident at lower [H]/[O] ratios, reflecting the effect of reaction (11) between 0 and H02 in producing maximum efficiency of consumption of 0 atoms per H02 formed in the reactor. Above [H]/[O] = 1, it is clear that only about one half of the maximum efficiency is achieved by the then dominant reactions (lo), (22) and (23) between H and H02, which reflects the production of OH radicals in reaction (1 0), subsequently reacting with 0 atoms in reaction (2).We have attempted to fit the form of fig. 1 using sets of values of kg, RH and Ro. In the first place the highest values of kg(app) suggest that k g cannot be less than - 1 x lo1' dm6 mok2 s-l. Moreover realistic extrapolation of the trend of lc,(app) with decreasing [H]/[O] suggests an upper limit of 1.4 x 1010 dm6 mok2 s-'. At the highest values of [H]/[O] shown in fig. 1, the range of kg(app) is encompassed by (6.0k0.5) x lo9 dm6 mo1-2 s-l, which will closely approximate to kg RH. RH is, therefore, indicated to be 0.5+0.1 by combination with the above range of kg.The two full curves in fig. 1 appear to offer reasonable fits to the data : curve (a) is produced with values of kg = 1.2 x 1O1O dm6 mok2 s-l, RH = 0.45 and Ro = 0.25, while curve (b) is based on kg = 1.05 x 1 O 1 O dm6 mol-1 s-I, RH = 0.50 and R, = 0.50. However, the two dashed curves do not appear to be able to fit the data adequately.2676 O+H,+02 SYSTEM' Curve (c) is based on k9 = 1.4 x lO1O dm6 in01-~ s-l, RH = 0.4 and Ro = 0.1, while curve (d) has k9 = 9.5 x lo9 dm6 mol-2 s-l, RH = 0.6 and Ro = 0.6. Both (c) and (d) fail to fit adequately at the left side of the figure, passing below the entire set of highest values of k,(app). Curves (a) and (b) indicate R, to be in the range 0.2 to 0.5 at 425 K, with the marginally better fit of (a) suggesting a median value of 0.3.[HI/[Ol FIG. 1.-Plot of k9(app) against [H3/[0] showing calculated curves which attempt to fit the variation. The full lines are considered acceptable, the dashed lines unacceptable; for the parameters used, see text. Kurylo and Wong and Davies have made independent measurements of kg(M = N2) at 298 K and 220 K or 226 K using flash-photolysis resonance-fluores- cence techniques. Their individual results are in extremely close agreement [l0-l0kg/ dm6 mo1-2 s-l = 1.92 and 1.92 (298 K) and 3.16 (226 K) and 3.03 (220 K)]. A straightforward Arrhenius extrapolation to 425 K predicts k9 = 1.1 x 1O1O dm6 mo1-2 s-l, which must lend strong support to our above analysis yielding k9 = (1.2k0.2) x 1O1O dm6 rn01-~ s-l.There is considerable uncertainty in the literature on both relative and absolute values of the rate constants for H atom reaction with HOz radicals, klo, kzz and k23, and only one measurement of the rate constant for 0 atom reaction with H02, kll. A recent evaluation has taken into account all available kinetic data to generate Arrhenius expressions for the rate constants ; the resulting predictions for 425 K are k,, = 2.5 x 1O'O (2), k,, = 1.1 x lo1* (2.5) and k,, = 1.5 x !OIO (>3) dm3 mol-l s-l, with the uncertainty factors given in the brackets. The middle values combineI . M. CAMPBELL, J . S . ROGERSON AND B . J . HANDY 2677 to produce RH - 0.33, in reasonable agreement with our result of 0.5 kO.1 considering the scale of uncertainty in the absolute values.The range of uncertainty in the literature relative results can be illustrated by the values for the ratio k22/(k10 +k23) of 0.51 k0.21 and 1.63k0.30 lo which were derived in two discharge flow studies at room temperature; the middle values yield upper limits (assuming k23 = 0) for RH of 0.67 and 0.38 respectively. of kl = (2.1 & 0.8) x 1O1O dm3 mol-l s-l at 293 K was obtained using laser magnetic resonance detection of HO, in a fast flow system. The magnitude of all these rate constants for H02 destruction suggests that their temperature dependences will be small. Simple combination of the middle values given above (ignoring the different temperatures) leads to a prediction of Ro N 0.4, uncertain to a factor of 2.5 and unlikely to vary by more than a factor of 2 if all the temperature coefficients could be taken into account.Thus our placing of R, in the range 0.2 to 0.5 at 425 K is quite reasonable and considerably improves our knowledge of this ratio parameter. At the same time since the rate constants concerned are not much less than the collision frequencies, it might be expected that Ro would be related to the inverse ratio of the square roots of the reduced masses of 0 + H02 and H+ H02 collision complexes. This is in fact 0.3 which may lend support to our measured value. The recent and only direct measurement T. M. Donahue, R. J. Cicerone, S. C. Liu and W. L. Chameides, Geophys. Res. Letters, 1976, 3, 105. I. M. Campbell and B. J. Handy, J.C.S. Farahy I, 1975,71,2097. I. M. Campbell and B. J. Handy, J.C.S. Faraday I, 1978,74, 316. I. M. Campbell and B. J. Handy, Chem. Phys. Letters, 1977, 47,475. N. Washida, R. I. Martinez and K. D. Bayes, 2. Naturforsch., 1974, 29a, 251. Chemical Kinetic and Photochemical Data for Modelling Atmospheric Chemistry, ed. R. F Hampson and D. Garvin (N.B.S. Technical Note), no. 866, 1975. M. J. Kurylo, .I. Phys. Chem., 1972, 76, 3518. W. Wong and D. D. Davis, cited in ref. (7) as private communication. M. A. A. Clyne and B. A. Thrush, Proc. Roy. SOC. A, 1963, 275, 559. l o A. A. Westenberg and N. de Haas, J. Phys. Chem., 1972, 76, 1586. l 1 J. P. Burrows, G. W. Harris and B. A. Thrush, Nature, 1977, 267, 233. (PAPER 8/696)